Are Different Modes of Weight Loss Associated with Survival in Oropharyngeal Cancer?

Abstract

Patients with oropharyngeal squamous cell carcinomas (OPSCC) often lose weight during radical radiotherapy (RT). Nutritional intervention is required in case of severe weight loss. However, weight loss does not provide full insight into body composition changes. Muscle mass is emerging as an important prognostic factor in cancer patients. We employed on-treatment cone-beam CT (CBCT) scans to monitor muscle mass and weight loss under the hypothesis that different modes of weight loss exist and may impact overall survival (OS). A retrospective analysis of 197 OPSCC patients treated with definitive or adjuvant (chemo)RT. Weekly weight measurements & CBCTs were collected. Patients were included if at least two time-points were available and the interval between the first and last was between 15-50 days. CBCTs were normalized to account for calibration differences between treatment machines. An in-house deep-learning model automatically segmented the skeletal muscle compartment at C3 on all CBCTs. Segmentations were visually checked and failures removed. Skeletal muscle volume (SMV, in mm3) was extracted after thresholding for intra-muscular fat. Relative changes in weight & SMV were then calculated. Linear models were fitted to each trajectory for every patient and slopes were estimated. The following weight & SMV categories were defined to generate equal groups: lost (more than 0.4 standard deviations (SDs) below the mean (M)), maintained (within +/- 0.4 SDs of M) or gained (more than 0.4 SDs above the M). Table 1 highlights the nine modes of body composition change. The prognostic value of these was investigated in multivariable Cox models accounting for age, sex, disease stage, oropharynx subsite, smoking status, performance status (PS), tumor p16 status, baseline weight & SMV, and treatment prescribed. The primary endpoint was OS. Mean weight & SMV changes during treatment were -0.047±0.001% & -0.044±0.019% per day. In multivariable analysis, gaining weight & losing SMV was identified as a significant risk factor for OS (p = 0.01, hazard ratio [95% CI]: 4.59 [1.40-15.10]). In this sub-group, mean weight & SMV change were +0.054±0.008% & -0.396±0.030% per day. PS>2 (p<0.001) & lower baseline weight (p = 0.02) were also significantly associated with OS. Patients losing substantial SMV but mildly gaining weight have significantly worse OS than others. This suggests there exists a group of patients where nutritional support is needed, but not offered because they maintain weight during treatment. Although our results need validation, continual monitoring of muscle condition during RT would allow these patients to be identified and promptly targeted for nutritional support.